Species review of the genus Boreophilia Benick from North America (Coleoptera, Staphylinidae, Aleocharinae, Athetini): Systematics, habitat, and distribution

Abstract Fourteen species of the genus Boreophilia Benick are now recognized in North America. Boreophiliainsecuta (Eppelsheim), reported by Lohse (1990) from North America, is a misidentification of a new species, which is described here as B.neoinsecuta Klimaszewski, sp. n., and the true B.insecuta (Epp.) does not occur in North America. An additional new species is found in Alaska, and described as B.beringi Klimaszewski & Brunke, sp. n. The following three species are synonymized (second name being valid): Boreophiliaherschelensis Klimaszewski & Godin, 2012, with Boreophiliavega (Fenyes, 1920); Boreophiliamanitobensis Lohse, 1990, with B.caseyi Lohse, 1990; and B.angusticornis (Bernahuer, 1907) with B.subplana (J Sahlberg, 1880), based on study of genital structures and external morphology. Athetagelida J Sahlberg, 1887, and Athetamunsteri Bernhauer, 1902, considered as Boreophilia in recent publications, are transferred to the genus Atheta Thomson, subgenus Dimetrota. Boreostibapiligera (J Sahlberg) is transferred to Boreophilia based on morphology and the results of our phylogenetic analysis. Boreophilianearctica is recorded from Alberta and B.nomensis is recorded from British Columbia for the first time. Each valid species is illustrated by color image of habitus, and black and white images of genitalia and tergite and sternite VIII. A new key to all Nearctic species of the genus is provided. DNA barcode data were available for nine of the 14 species, which we downloaded, analyzed, and used as additional evidence for the taxonomic conclusions reached herein.


Introduction
Boreophilia G Benick, 1973, is a small athetine genus, comprising Nordic species distributed exclusively in the Palaearctic and Nearctic regions. There are 17 species recorded in the Palaearctic (Smetana 2004), and 14 in the Nearctic regions. Of these, six species have a circumpolar Holarctic distribution, but two species included by Smetana (2004) in this genus, B. gelida (J Sahlberg) and B. munsteri (Bernhauer), are here transferred back to the genus Atheta where they were originally described, on the grounds of morphology of genital characters, including sexual modification of male tergite VIII, which is unmodified in Boreophilia. Consequently, there are 15 Palaearctic and 14 Nearctic species, of which six are Holarctic. It is interesting that none of the species with elytra shorter than the pronotum (and presumably brachypterous), were shown to be Holarctic. Limited dispersal in these groups has likely led to allopatric speciation between Nearctic and Palaearctic populations (e.g., B. piligera and B. beringi sp. n.). The Holarctic species constitute ca. 24% of the total fauna of the genus, which is likely the highest percentage of Holarctic species among Nearctic aleocharine genera. This genus, like Gnypeta CG Thomson, another northern Holarctic genus, is a good target for monitoring climate warming and its effects on distribution and survival of Nordic species. We here update our knowledge on all recorded Nearctic species and synonymize four species from previous records . Two Nearctic species are described as new. All Nearctic species are classified to species groups, which presumably reflect their close relationships.
In the past, there was confusion regarding some Nearctic species of Boreophilia because species of this genus have similar structures of the median lobe of the aedeagus and of the spermatheca, insufficient material was available for study, and a general poor knowledge of Palaearctic species in the Nearctic region. We have corrected these as much as the available material permitted and have provided better diagnoses for Nearctic species. We have also studied European material to compare with Nearctic specimens of selected Holarctic species. This resulted in additional synonymy and clarification as to the known distribution of many species in North America. Brundin's (1954) paper was very helpful to our study. This work was clearly ahead of its time, with perfectly accurate illustrations of the median lobe of the aedeagus and spermatheca of several European species, providing the best diagnostic characters at the species level. We hope to encourage other scientists to use species of this genus in monitoring the effects of climate change on species in Nordic environments.

Material and methods
Almost all specimens used in this study were dissected, and their genital structures examined. The genital structures were dehydrated in absolute ethanol and mounted in DNA barcode data were downloaded from the BOLD website (http://www. boldsystems.org) after applying filters to exclude those flagged as misidentifications, those with sequence lengths under 100 bp, those with stop codons, and those flagged as contaminated. This resulted in sequence data for nine of the 14 species included herein. The amino acid based HMM BOLD aligner was used to align the data prior to download. Two sequences each of Atheta cinnamoptera and Atheta munsteri were used as outgroups. The latter species was also included to test its generic placement. This resulted in a dataset of 33 sequences. Of 654 base pairs in the alignment, 455 are constant, 19 are variable but parsimony uninformative, and 180 are parsimony-informative. Specimens of all included Boreophilia were identified to species via morphological study, or to genus for some females. These sequences came from a variety of projects (Table 1) and publications (Elven et al. 2010, Pentinsaari et al. 2014, Sikes et al. 2017. The NEXUS file with the alignment and resulting tree is available for download from https://doi.org/10.6084/m9.figshare.7822496. To obtain a robust estimate of the mtDNA gene tree using these DNA barcode data, PartitionFinder2 (Lanfear et al. 2016) was used via the CIPRES Science Gateway (Miller et al. 2010) to obtain the best partitioning and modeling scheme. We used the following parameters for the cfg file: alignment = infile.phy, branchlengths = linked, models = all, model selection = aicc, search = greedy, with each codon position indicated as a separate partition. Mesquite v3.6 (Maddison and Maddison 2018), was used to export the original Nexus file to Phylip format for PartitionFinder. The best scheme chosen by PartitionFinder retained each codon position as a partition with first codon positions modeled using the TrN+I+G model, second positions modeled using the F81+I model, and third positions modeled using the GTR+G model. All DNA distances reported herein are uncorrected, p-distances. Minimum, average, and maximum distances were calculated in Excel from a distance matrix generated by PAUP 4.0a (build 164) (Swofford 2002). This data file is available at https://doi.org/10.6084/m9.figshare.7822508.
Bayesian and maximum likelihood phylogenetic analyses were conducted via the CIPRES portal using MrBayes v3.2.6 without the BEAGLE option (Ronquist et al. 2012) and Garli 2.0 (Zwickl 2006). Because MrBayes doesn't have the TrN model, for the first codon position we used the GTR model, which PartitionFinder selected for use with MrBayes. Two runs of four chains each were sampled for 8 million generations with samples taken every 1000 generations; the first 25% of the samples were discarded as burn-in, yielding 12,002 samples. The average standard deviation of the split frequencies was 0.003158 and the average Potential Scale Reduction Factor (Gelman and Rubin 1992a, b) was 1.000, thus indicating convergence had been reached. The sampling was considered adequate based on the average estimated sample sizes (ESS) of the parameters all being greater than 2000, as assessed by MrBayes. Also using the CIPRES portal, we ran 200 bootstrap replicates composed of four search replicates each using GARLI, with zero length branches collapsed. The resulting trees were imported into PAUP 4.0a (build 164) (Swofford 2002) to produce a 50% majority rule consensus tree, the node support values of which were transferred to the Bayesian consensus tree. An additional maximum likelihood analysis was conducted in IQTREE 1.6 (Nguyen et al. 2015) on an iMac (4 GHz i7, 16GB) to acquire alterna-tive node support values, namely the ultrafast bootstrap of Hoang et al. (2017) and the SH-aLRT test of Guindon et al. (2010). The analysis was performed using the same partitioning scheme as used for GARLI and with the -spp option, which allows partition-specific rates, 500 search replicates, and other parameters set to defaults. Clade support was assessed using 1000 replicates of the ultrafast bootstrap and an SH-aLRT test with 1000 replicates. Nodes with support values of both UFB ≥ 95 and SH-aLRT ≥ 80 are considered well supported (Nguyen et al. 2015), nodes with one of UFB < 95 or SH-aLRT < 80 are considered weakly supported, and nodes with both UFB < 95 or SH-aLRT < 80 are considered unsupported.

Phylogenetic results
The resulting estimate of the mtDNA gene tree ( Fig. 1) was relatively well resolved although a few relationships were obscured by polytomies or ambiguous due to low branch support values. The genus Boreophilia, as defined herein via morphology, was strongly supported as monophyletic (PP = 1.0, BS = 100, UFB = 100, SH-aLRT = 100). Notably, the species we transferred to Boreophilia (B. piligera) from Boreostiba was recovered within the clade of other Boreophilia while a species we transferred out of Boreophilia and into Atheta (A. munsteri) was recovered as the closest lineage to Boreophilia, with a long branch separating the two clades, thus supporting its exclusion from Boreophilia. Zero of the sampled Bayesian trees had A. munsteri nested within the Boreophilia clade, thus failing to reject the hypothesis that it is not a Boreophilia as morphologically defined herein. The fusca species group was supported as monophyletic with a strong posterior probability (0.98), ultrafast bootstrap (95%) and SH-aLRT support (82%) but relatively weak maximum likelihood bootstrap support (63%). The subplana species group, however, was not recovered as monophyletic due to its members and the fusca group emerging from a polytomy. Given the small size of the dataset, the subplana species group hypothesis remains ambiguous. All species with multiple specimens sampled were recovered as monophyletic with strong support (PP = 0.99 -1.0, BS = 78-99% UFB = 84-100%, SH-aLRT = 89-100%) including two species, B. hyperborea and B. fusca, with samples from both the Nearctic and Palearctic. Nine of the ten species in our analysis are in BINs on BOLD (Table 1) with no species occurring in more than one BIN, and with no BIN holding more than one morphologically identified species.
Given the relatively small size of the dataset, in both taxon sampling and genetic data, we refrain from drawing any biogeographic conclusions based on these preliminary phylogenetic analyses. Additional genes including nuclear markers, greater specimen sampling within species, and addition of the missing Boreophilia species, would greatly improve our understanding of the evolution of these taxa.
Five specimens were female and could not be identified with certainty based on morphology alone (Fig. 1). Two of these (B. sp. TWCOL345_09 from Manitoba, Canada, and B. sp. COLFC286_12 from Finland) show branch lengths large enough to potentially be unique species not already represented in our dataset. The remaining Figure 1. Fifty percent majority rule consensus phylogram from the Bayesian analysis with branch support values provided from left to right as: estimated posterior probabilities, maximum likelihood bootstrap proportions, ultrafast bootstrap values, and an SH-aLRT test values, with * = bootstrap values below 50%. Taxon identity is indicated for each sequence, followed by abbreviations of locality, and BOLD process IDs (see Table 1

Summary of DNA distances within and among species
A full spreadsheet of DNA distances and our calculations is archived at https://doi. org/10.6084/m9.figshare.7822508. We summarize the key findings here. Among Boreophilia species, the minimum uncorrected 'p' distance within a species (limited only to sequences identified to species via morphology) in our dataset was 0.00%, the mean within species distance was 0.280%, and the maximum within species distance was 1.072%. Surprisingly, this maximum distance was found between two Nearctic samples (B. nomensis from British Columbia versus B. nomensis from Alaska) rather than between Palearctic versus Nearctic conspecific samples (B. fusca from Finland versus B. fusca from Manitoba, Canada were 0.539% distant and B. hyperborea from Russia versus B. hyperborea from Manitoba, Canada were 0.155% distant). This maximum within species distance is not a result of one of these sequences being incomplete (the British Columbia sequence is only 564 bp long while the Alaska sequence is 658 bp long). When these two sequences were compared after excluding base pairs missing from the shorter sequence, so both were 564 bp long, their distance was 1.064%, which remains the maximum within species value.
The median lobe of aedeagus of B. eremita is similar to that of B. islandica, but tubus is more elongate, narrower, less arcuate, and with basal projection angular in lateral view (Figs 7,8), in dorsal view tubus with two pointed latero-basal projections (Figs 9,10). In B. islandica, the tubus is broader, shorter, more arcuate ventrally, and with basal arcuate projection rounded in lateral view (Fig. 17), in dorsal view tubus with two smaller and less angular basal projections (Fig. 18). Spermathecae of the two species is variable in shape and very similar, female tergite and sternite VIII are similar in shape but sternite VIII in B. eremita has feebly arcuate medial part of antecostal suture (Fig. 14), which is strongly sinuate in B. islandica in the majority of examined specimens (Fig. 22). On average, the body of B. eremita is narrower and elytra shorter than that of B. islandica.
Distribution. Holarctic species; recorded from north and central Europe, Ireland, Ukraine, Russia (west and east Siberia) and the Russian Far East; Canada: LB, NB, MB; USA: AK.
Collection data. Habitat: in NB -old silver maple forest with green ash and seasonally flooded marsh; silver maple swamp, margin of vernal pond, found in moist leaves. In AK -creekside/ocean beach confluence, under boards and drift wood; black and white spruce, willow; subalpine habitat with Veratrum, and Calamagrositis. Comments. We have examined several European specimens identified as B. hercynica, which have the shape of median lobe of aedeagus and spermatheca similar to B. eremita, but the body color and the shape and proportions of forebody were different: body brown with dark brown head and pronotum and particularly elytra paler, and pronotum strongly transverse with sides broadly and evenly arcuate, elytra at suture slightly shorter than pronotum along midline. These specimens may represent extreme variation of B. eremita or a different and distinct species. Additional material is needed, and possibly DNA studies, to establish clear status of these specimens.  Diagnosis. Body broad, forebody moderately and abdomen strongly glossy (Fig. 16); length 2.8-3.5 mm; uniformly black with paler, reddish brown appendages, or head, pronotum and VI-VII basal segments of abdomen dark brown, remainder of the body reddish brown, appendages light brown, sometimes elytra with some reddish tinge; antennomeres VIII-X subquadrate; pronotum as long as elytra at suture, maximum width of pronotum distinctly less than maximum width of elytra; elytral length variable, as long as pronotum or slightly longer. Male. Tubus of median lobe of aedeagus with two basolateral and slightly angular apically projections in dorsal view (Fig. 18), and one rounded projection in lateral view (Fig. 17); bulbus moderately broadly oval with two elongate sclerites of internal sac in dorsal view (Fig. 18); tergite VIII arcuate apically (Fig. 19); sternite VIII elongate, parabolic apically (Fig. 20). Female. Spermatheca: capsule pitcher-shaped basally with tubular apical projection moderately long and narrowed apically, stem coiled posteriorly, there is great variability in the shape of capsule (Fig. 23); tergite VIII arcuate apically (Fig. 21); sternite VIII rounded apically, antecostal suture strongly sinuate in most examined specimens (Fig. 22) Comments. Females of this species may be confused with other species of Boreophilia and particularly those of closely related B. eremita. Associating females with males is considered here to be the most reliable way of identifying females of this and the previous species. At present, B. islandica is considered a somewhat variable species. Specimens vary from moderately robust and narrower to more robust and broader, with elytra as long as pronotum or slightly longer, all with the same morphology of genitalia. The BIN BOLD:ABX3767 formed a sister group to B. islandica in our analysis and was represented by a single female from Finland. The capsule of its spermatheca is curved at an angle of nearly 90 degrees and was among those shapes included in the illustrations of B. islandica by Palm (1970). BIN BOLD:ABX3767 may represent an undescribed species, or one of the described species not included in our DNA dataset, but corresponding males should be examined. A separate DNA based study is recommended to examine specimens of B. islandica sensu lato from a broad Holarctic distribution, including males and females.
Additional material examined. NEARCTIC: Canada, Newfoundland, Long Range Mts., Portland Cr. Hill, 12-13.VIII.1982  DNA Barcode data. Our data included one sequence identified as B. islandica from Newfoundland and Labrador, Canada, and one sequence identified as Boreophilia sp. collected from Churchill, Manitoba, which are the only members of BIN BOLD:AAH0226. BOLD reports these sequences are 2.79% distant from their nearest neighbor.
DNA Barcode data. Our data included five sequences of specimens identified as B. nearctica, two from Alaska and three from Alberta, Canada, which grouped with two sequences identified as Boreophilia sp. into BIN BOLD:ACU9385. Our calculations indicate that the five sequences identified to species have an average distance of 0.14%, a maximum distance of 0.33% and are 6.37% distant from their nearest neighbor.

Boreophilia ovalis Klimaszewski & Langor, 2011 Figs 53-56
Boreophilia ovalis Klimaszewski & Langor, in Klimaszewski et al. 2011: 186 Diagnosis. Body very broad, forebody moderately and abdomen strongly glossy (Fig.  53); length 3.0-3.5 mm; head, pronotum and abdomen except for its apex dark brown, elytra dark-reddish brown medially, appendages brown, or entire body dark brown to almost black and tarsi reddish brown; antennomeres VIII-X elongate; pronotum ca. as long as elytra at suture, maximum width of pronotum slightly less that the maximum width of elytra. Male. Unknown. Female. Spermatheca: capsule pitcher-shaped basally with broadly tubular and slightly pointed apical part, stem short, strongly sinuate and looped posteriorly (Fig. 56); tergite VIII arcuate apically (Fig. 54); sternite VIII rounded apically, antecostal suture strongly sinuate medially (Fig. 55). Females of this species may be confused with those of B. fusca, from which they differ by distinctly elongate antennomeres VIII-X, more deeply medially sinuate antecostal suture of sternite VIII, and spermathecal capsule more evenly elongate and apex less pointed laterad. Distribution. Nearctic species, recorded only from Canada: NF. Collection data. Habitat: unspecified forest. Collecting methods: one female was captured in Malaise trap. Collecting period: June to September.
DNA Barcode data. Two specimens of this species, one being a paratype, were submitted for DNA barcoding but failed to generate DNA sequences (process IDs on BOLD: LFCAB222-15, NGSFT931-15).
Diagnosis. Body narrow, subparallel, moderately glossy, abdomen slightly more so (Fig.  57); length 3.0-3.8 mm; head, pronotum and abdomen dark brown, elytra reddish brown, legs paler, yellowish brown; antennomeres VIII-X subquadrate; pronotum longer than elytra at suture, maximum width of pronotum ca. the same as maximum width of elytra. Male. Tubus of median lobe of aedeagus almost straight (slightly arcuate) in lateral view, apex triangular in shape, slightly pointed (Fig. 58), bulbus broad in dorsal view and with two elongate narrow sclerites as illustrated (Figs 59, 60); tergite VIII arcuate apically (Fig. 61); sternite VIII elongate, parabolic apically (Fig. 62).  DNA Barcode data. Our data included two sequences of specimens identified as B. nomensis, one from Alaska and one from British Columbia, Canada which grouped into BIN BOLD:ACU9384. BOLD reports these sequences have an average and maximum distance of 1.06% and are 4.64% distant from their nearest neighbor.
Distribution. Holarctic species, recorded from Europe, Finland; Asia, East and West Siberia, Mongolia; and North America: Canada: YT; USA: AK [new record]. Comments. , described Dimetrota venti from Yukon. This species is extremely similar externally and genitally to Boreophilia insecuta described from Europe. The genitalia of B. insecuta were illustrated by Brundin (1954), and here, based on a specimen from Siberia (Figs 82, 83).
In B. venti, the tubus of the median lobe is distinctly arcuate in lateral view (Fig.  64), but in B. insecuta it is always straight basally (Fig. 82) and the two main sclerites are slightly different in shape (Figs 64,65,82,83). However, the remaining genital characters are similar and females of the two species are not distinguishable morphologically. Etymology. Derived from prefix neo-added to existing specific name insecuta, a closely related species.
Distribution. Nearctic, Canada: MB, YT: USA: AK. Comments.  reported Boreophilia insecuta (Eppelsheim) in North America from AK, MB, YT, as a Holarctic species. However, these records represented a different and undescribed Nearctic species, which is here described as B. ne-oinsecuta sp. n. The two species, B. insecuta and B. neoinsecuta, have similarly shaped genitalia, which explains why they were confused. The true B. insecuta (Eppelsheim) was illustrated by Brundin (1954), and here (Figs 82,83), and has a distinctly dilated apex of median lobe of aedeagus in lateral view (Fig. 82), while it is narrower in B. neoinsecuta (Figs 72,73), and the two sclerites of internal sac of median lobe of aedeagus are broader and differently shaped than those of B. neoinsecuta (Figs 82, 83). Spermathecae of both species are very similar in shape, but female sternite VIII in B. insecuta is apically rounded and truncate medially, while in B. neoinsecuta is triangularly produced apically and pointed medially (Fig. 78).
DNA Barcode data. Our data included two sequences of B. neoinsecuta paratypes, both from Alaska which grouped into BIN BOLD:ADR7545. These sequences are 0.00% distant from each other and BOLD reports they are 7.23% distant from their nearest neighbor. Etymology. Named after Danish explorer Vitus Bering, whose name is shared with the species' type locality, Bering Land Bridge National Park, and 'Beringia', the area of adjacent Russia and Alaska that were previously connected multiple times during the past 1 million years.
Comments. We here compared Palaearctic Boreostiba piligera (J. Sahlberg), two males from Finland (ZMUO, NHMD), with our new species from Alaska. The two species are very similar externally but may be distinguished by the different morphology of the median lobe of aedeagus. In B. piligera, the apical part of the tubus is more weakly deflexed ventrad and distinctly more elongate. The sclerites of the internal sac  are also more elongate and quite differently shaped (less like a talon and more even in thickness along their length). Therefore we conclude that these two populations represent sibling species.
Diagnosis. Body moderately broad, strongly glossy, abdomen slightly more so (Fig.  92); length 2.8-3.5 mm; black, elytra with some reddish tinge, tarsi yellowish brown; antennomeres VIII-X slightly elongate; pronotum shorter than elytra at suture, maximum width of pronotum distinctly less than maximum width of elytra. Male. Tubus of median lobe of aedeagus almost straight and slightly produced ventrad in lateral view, apex broad and rounded (Fig. 93), bulbus broad and with two elongate sclerites of distinctive shape (Figs 93, 94); tergite VIII truncate or slightly concave apically (Fig. 95); sternite VIII elongate, parabolic apically (Fig. 96). Female. Spermatheca: capsule broadly club-shaped, moderately long and rounded apically, with small apical invagination; stem sinuate, straight medially, and looped posteriorly (Fig. 99); tergite VIII arcuate apically (Fig. 97); sternite VIII triangularly produced apically and pointed medially, antecostal suture arcuate, straight medially (Fig. 98 Comments. Bernhauer (1907) described Atheta (Metaxya) angusticornis from Mount Washington, New Hampshire, USA. Gusarov (2003) transferred it to the genus Boreophilia, and considered it closely related to B. subplana, from which he differentiated it by the "shape of aedeagus, particularly a narrower apex of the median lobe in parameral view". We have studied the median lobe of AK and NH specimens and found no differ-ences warranting different species recognition. We therefore consider the two populations as belonging to the same species. The Mount Washington, NH, population represents a remnant, southernmost population of this species known only from higher elevations.
DNA Barcode data. Our data included two sequences of B. subplana, both from Finland but because they are < 500 bp in length, they were not assigned to a BIN on BOLD. Our calculations indicate these sequences have an average and within-species maximum distance of 0.0% and are 6.37% distant from their nearest neighbor.  Figs 100-108
Comments.  described the new species B. manitobensis from MB and AK. The holotype from MB is represented by a male with a distorted median lobe of the aedeagus. We have studied the external and internal morphology of the two species, including the structures of internal sac, and found no significant differences between B. manitobensis and B. caseyi. Therefore, B. manitobensis is here synonymized with B. caseyi. The two species were published in the same paper, but B. caseyi has page priority and therefore was chosen as a valid species.
DNA Barcode data. Four specimens of B. caseyi from UAM were submitted for DNA barcoding and three did not produce DNA sequences. The one which was successfully sequenced was flagged on BOLD as possibly contaminated so we excluded it from our analyses.
Collection data. Habitat: Yukon specimens were collected in an alluvial fan in June and July .
Comments. Boreophilia herschelensis is conspecific with B. vega and is here synonymized. Boreophilia vega has a median lobe of aedeagus similar to that of B. neoinsecuta (Fig. 72) and B. insecuta (Fig. 82), but the spermatheca of B. vega compared to the latter two species is differently shaped (Figs 79, 80, 116). Externally, B. vega differs from B. neoinsecuta in elytra distinctly broader that the maximum width of pronotum (Figs 71, 109). Female sternite VIII is differently shaped in the two species (Figs 78, 115).
DNA Barcode data. Our data included one sequence of B. vega from Yukon Territory, Canada, but because this sequence was < 500 bp long it was not assigned a BIN on BOLD. We calculate that this sequence is 6.5 % distant from its nearest neighbor.
This species may be distinguished by the unique shape of spermatheca. Distribution. Nearctic species, known only from Canada, YT. Collection data. Habitat: white spruce and feathermoss forest, mixed pine and willow forest, black spruce stand, mixed aspen and white spruce forest . Females from Yukon were collected in May to September using pitfall traps and litter sifting in mature forest.
Comments. This species is tentatively assigned to this group, because the male is unknown and morphology of median lobe of aedeagus could not be analysed.